Rate Constants for CH3 + O2 → CH3O + O at High Temperature and

[CH3I]0 was sufficiently low so that most secondary reactions were negligible, allowing for unambiguous determination of the rate constant for CH3 + O...
0 downloads 0 Views 80KB Size
5942

J. Phys. Chem. A 1999, 103, 5942-5948

Rate Constants for CH3 + O2 f CH3O + O at High Temperature and Evidence for H2CO + O2 f HCO + HO2 J. V. Michael,* S. S. Kumaran,† and M.-C. Su‡ Chemistry DiVision, Argonne National Laboratory, Argonne, Illinois 60439 ReceiVed: March 17, 1999; In Final Form: May 17, 1999

The reaction of CH3 with O2 has been studied in a reflected shock tube apparatus between 1600 and 2100 K. CH3 was prepared from the fast thermal decomposition of CH3I, and O atom atomic resonance absorption spectrometry (ARAS) was used to observe absolute [O]t. [CH3I]0 was sufficiently low so that most secondary reactions were negligible, allowing for unambiguous determination of the rate constant for CH3 + O2 f CH3O + O. The rate constant expression for this reaction derived to match the experimental data is k ) (3.90 ( 0.40) × 10-11 exp(-16858 ( 1127 K/T) cm3 molecule-1 s-1. To explain O atom concentration buildup at longer times, a fast reaction between H2CO and O2 was postulated. Rate constants for this process were derived by fitting the long-time O atom profiles. Last, the four-center reaction CH3 + O2 f CH2O + OH was found to be of negligible importance over the temperature range of the study.

Introduction The reaction of CH3 with O2 is one of the most important reactions in hydrocarbon combustion. This issue has of course been recognized for over 50 years, and the work on the reaction is extensive as shown by the fact that there are 107 references to it in the NIST database.1 Three elementary reactions are possible:

CH3 + O2 (+M) f CH3O2 (+M)

(1)

f CH2O + OH

(2)

f CH3O + O

(3)

Reactions 2 and 3 are forward dissociation channels from the initially formed vibrationally hot methylperoxy radical in competition with stabilization (reaction 1). As temperature increases, the back process, reaction -1, also becomes important in an RRKM sense, decreasing the overall rate of reaction 1 to negligible values above ∼1500 K and below ∼400 Torr. On the other hand, reactions 2 and 3 have activation barriers, and these channels open only at higher temperatures and dominate reaction 1. Hence, under the present high-temperature conditions, reactions 2 and 3 are the important reactions. The branching ratio of reactions 2 and 3 has been a perplexing problem for a long time. Recently, Yu, Wang, and Frenklach2 have thoroughly discussed the earlier data on both reactions and have noted that there is agreement between the various values for reaction 3 (within a factor of 3), but the spread in values for reaction 2 is over an order of magnitude. These workers shock-heated methane-oxygen mixtures and spectroscopically measured both [OH]t and [CO]t. They then used * To whom correspondence should be addressed. Address: D-193, Bldg. 200, Argonne National Laboratory Argonne, IL 60439. Phone: (630) 2523171. Fax: (630) 252-4470. E-mail: [email protected]. † Present address: Cabot Corporation, 700 E. US Highway 36, Tuscola, IL 61953. ‡ Faculty Research Participant, Department of Educational Programs, Argonne. Permanent address: Department of Chemistry, Butler University, Indianapolis, IN 46208.

optimization techniques and GRI-Mech3 to fit the data, the most important CH3 oxidation rate processes being reactions 2 and 3. They report

k2 ) 3.07 × 10-12 exp(-10224 K/T) cm3 molecule-1 s-1 (4) and

k3 ) 4.90 × 10-11 exp(-15340 K/T) cm3 molecule-1 s-1 (5) For reaction 2, they showed that the earlier data could be separated into two distinct sets, one suggesting high values4-11 and the other suggesting low values for k2.12-19 Their result, eq 4, clearly suggests that the low values are preferred. With regard to reaction 3, the results were more uniform. Except for the data of Saito et al.,10 the earlier values11,14-17,20,21 agreed with eq 5 to within a factor of 2. The results for reaction 3 would then seem to be consistent; however, in a recent report,19 BraunUnkhoff et al. have indicated that their [OH]t measurements required k3 to be about one-half to one-third of eq 5. Similarly, OH and CH3 profile results by Tischer et al.22 have given values for k3 that are lower than eq 5 by about one-half. Hence, we have concluded that the situation is still perplexing, and this realization has prompted the present study. In the present work, the formation rates of O atoms are measured behind reflected shock waves using the atomic resonance absorption spectrometry (ARAS) technique. The experiments are similar to those of Bhaskaran et al.14 and Klatt et al.17 except that O atom detection is about 5 times more sensitive than in these two earlier studies. This is important because the complications from secondary reactions can be suppressed by observing O atom production at lower reactant concentrations. Experimental Section The present experiments were performed with previously described equipment,23 and therefore, only a brief description

10.1021/jp9909457 CCC: $18.00 © 1999 American Chemical Society Published on Web 07/09/1999

CH3 + O2 Reaction

J. Phys. Chem. A, Vol. 103, No. 30, 1999 5943 TABLE 1: Mechanism Used for Determining (ABS) against [O] from the H + O2 Reaction Using C2H5I as a Source of H Atomsa 1. C2H5I (+M) f C2H4 + H + I (+M), k1 ) 6.34 × 109 exp(-15894 K/T)b 2. C2H5I (+M) f C2H4 + HI (+M), k2 ) 0.15 × k1b 3. H + O2 f OH + O, k3 ) 1.62 × 10-10 exp(-7474 K/T)c 4. OH + OH f O + H2O, k4 ) 7.19 × 10-21T2.7 exp (1251 K/T)c 5. H + HI f H2 + I, k5 ) 7.87 × 10-11 exp(-330 K/T)d 6. I + H2 f HI + H, k6 ) 2.81 × 10-10 exp(-16930 K/T)d a All rate constants are in cm3 molecule-1 s-1 except for reactions 1 and 2, which are in s-1. b Reference 28. c Reference 30. d Reference 31.

TABLE 2: Data for O Atom Curve-of-Growth Determination P1/ Torr

Msa

F5b/ (1018 cm-3) T5b/K

[C2H5I]0c/ cm-3

XC2H5I ) 6.266 × 10 , XO2 ) 1.763 × 10 2.580 2.163 1656 c1.356 (13) 1.211 (13) 2.618 1.772 1702 1.110 (13) 1.010 (13) 2.623 1.602 1707 1.004 (13) 0.902 (13) 2.555 1.376 1627 0.862 (13) 0.789 (13) 2.589 1.208 1667 0.757 (13) 0.679 (13) 2.612 1.002 1694 0.628 (13) 0.569 (13) 2.600 3.147 1673 1.972 (13) 1.769 (13) 2.489 3.035 1547 1.902 (13) 1.694 (13) 2.431 2.046 1485 1.282 (13) 1.180 (13) 2.417 1.819 1470 1.140 (13) 1.036 (13) 2.474 1.639 1534 1.027 (13) 0.916 (13) 2.394 1.375 1445 0.862 (13) 0.767 (13) 2.460 1.430 1519 0.896 (13) 0.787 (13) 2.594 1.302 1673 0.816 (13) 0.684 (13) -6

Figure 1. Curve-of-growth determination for O atom ARAS: (9) from ref 25 [O] data based on the unimolecular decomposition of N2O; (b) [O] derived from a simulation using the mechanism given in Table 1. (ABS) is the experimentally determined absorbance (see text).

of the system, along with those features unique to the current experimental procedures, will be presented here. Apparatus. The apparatus consists of a 7 m (4 in. o.d.) 304 stainless steel tube separated from the He driver chamber by a 4 mil unscored 1100-H18 aluminum diaphragm. The tube was routinely pumped between experiments to less than 10-8 Torr by an Edwards Vacuum Products model CR100P packaged pumping system. The velocity of the shock wave was measured with eight equally spaced pressure transducers (PCB Piezotronics, Inc., model 113A21) mounted along the downstream part of the test section of the shock tube and recorded with a 4094C Nicolet digital oscilloscope. Temperature and density in the reflected shock wave regime were calculated from this velocity. This procedure has been given previously, and corrections for boundary layer perturbations have been applied.23,24 The digital oscilloscope was triggered by pulses derived from the last velocity gauge signal. The photometer system was radially located 6 cm from the end plate. All optics were made from MgF2. The resonance lamp beam intensity was measured by an EMR G14 solar blind photomultiplier tube and recorded with the oscilloscope. O Atom ARAS. The technique used for the detection of the transient O atoms formed in reaction 3 is atomic resonance absorption spectroscopy (ARAS). In earlier work,25 O atom curves-of-growth were determined for two different resonance lamp conditions: (1) XO2 ) 1 × 10-3 in 1.8 Torr of purified grade He and (2) no added O2 in purified grade He, both operating at 50 W microwave power to give an effective lamp temperature of 490 K.26 Method 2 relies on the O2 impurity in the He tank and has a less reversed line shape than method 1; however, method 1 was used in the present work because the signal-to-noise ratio is better. The nonoverlapping 130.2, 130.4, and 130.6 nm triplet resonance lines are isolated with a CaF2 window (transmitting at wavelengths greater than 125 nm), and the fraction of the light that is resonance radiation is determined by using an O atom atomic filter section in front of the lamp as described previously.25 In the earlier work,25 O atom concentration calibration was carried out using the thermal decomposition of N2O. These measurements are shown as solid squares in Figure 1. From these data alone the derived curve-of-growth (COG), calculated from atomic resonance line absorption theory,25-27 is reproduced

10.94 8.84 7.98 7.02 6.09 5.01 15.90 15.95 10.97 9.81 8.63 7.49 7.57 6.55

[O]simc/ cm-3

(ABS)

-3

0.913 0.755 0.688 0.623 0.545 0.481 1.154 1.096 0.801 0.720 0.695 0.584 0.640 0.574

a The error in measuring the Mach number, M , is typically 0.5s 1.0% at the one standard deviation level. b Quantities with the subscript 5 refer to the thermodynamic state of the gas in the reflected shock region. c Parentheses denote the power of 10.

as the dashed line in Figure 1. We found that the COG for the moderately reversed source was not strongly dependent on temperature, as shown by the data and line absorption calculations. In the present work, the relationship between [O] and absorbance (ABS) ((ABS) ≡ -ln(I/I0) where I0 and I refer to initial and transmitted resonance light intensities, respectively) was determined under slightly lower temperature conditions than before.25 The present method utilizes the reaction

H + O2 f OH + O

(6)

at low [H]0 where the H atoms are produced from the fast thermal decomposition of C2H5I.28,29 The pertinent mechanism is given in Table 1 along with well-established rate constants for all processes. Experiments have been performed by varying [C2H5I]0 in excess O2, and absorption profiles were measured at several temperatures. The absorption profiles rapidly decreased to a near-constant level. [O] was then simulated over the time necessary to obtain near-constant absorption by numerically integrating the mechanism of Table 1. The conditions of the experiments are shown in Table 2. For 14 experiments in the table, the simulated [O] is listed along with the measured (ABS), and these values are subsequently plotted in Figure 1 (closed circles) along with the earlier results from the N2O decomposition study.25 It should be noted that reactions 5 and 6 in Table 1 are negligible, contributing less than 0.5% to [O]. The profile is mostly affected by (a) the branching ratio of reactions 1 and 2 and (b) by the H + O2 reaction. The maximum contribution to [O] from the OH + OH (reaction 4 in Table 1) at the highest temperature and [C2H5I]0 is less than 5%.

5944 J. Phys. Chem. A, Vol. 103, No. 30, 1999

Michael et al.

TABLE 3: High-Temperature Rate Data for CH3 Oxidation by O2 P1/ Torr

Msa

F5b/ (1018 cm-3)

T5b/K

k3c,d/ (cm3 s-1)

k16c,d/ (cm3 s-1)

5.92 5.96 5.92 7.48 7.46

XCH3I ) 2.482 × 10-6, XO2 ) 1.056 × 10-1 2.682 1.333 1644 1.95 (-15) 2.823 1.410 1802 3.00 (-15) 3.055 1.502 2078 1.20 (-14) 2.959 1.847 1962 8.00 (-15) 3.045 1.887 2066 1.04 (-14)

2.0 (-15) 3.0 (-15) 8.0 (-14) 2.0 (-14) 5.0 (-14)

10.96 10.79 10.95 10.99 10.93 10.91 10.96 10.97 10.99 10.98 10.95 10.93 10.93

XCH3I ) 1.592 × 10-6, XO2 ) 5.822 × 10-2 2.865 2.517 1916 6.50 (-15) 2.998 2.574 2083 1.70 (-14) 3.018 2.626 2109 1.60 (-14) 2.960 2.594 2035 1.00 (-14) 2.824 2.479 1867 4.00 (-15) 2.688 2.357 1712 1.65 (-15) 2.597 2.290 1608 1.40 (-15) 2.715 2.392 1744 2.20 (-15) 2.647 2.340 1665 1.50 (-15) 2.776 2.444 1815 2.20 (-15) 2.909 2.539 1977 9.50 (-15) 2.971 2.579 2055 1.00 (-14) 2.999 2.599 2091 9.00 (-15)

1.5 (-14) 7.0 (-14) 7.0 (-14) 5.0 (-14) 1.5 (-14) 4.0 (-15) 1.0 (-15) 4.0 (-15) 2.0 (-15) 5.0 (-15) 1.5 (-14) 6.0 (-14) 5.0 (-14)

a The error in measuring the Mach number, M , is typically 0.5s 1.0% at the one standard deviation level. b Quantities with the subscript 5 refer to the thermodynamic state of the gas in the reflected shock region. c Fitted rate constants for reactions 3 and 16 of Table 4 (see text). d Parentheses denote the power of 10.

From Figure 1, the earlier data from the N2O decomposition agree with the results based on the Table 1 mechanism within experimental error. We showed earlier that our H atom ARAS calibration methods26 agreed with independent Cl 32 and I atom33 ARAS results. The present results are in fact a calibration between H and O atom ARAS methods and, as a result, corroborate the Table 1 mechanism and, particularly, the branching ratio between reactions 1 and 2. Last, we have parametrically applied line absorption calculations25-27 to the composite set (the present and earlier data25), allowing only [O] in the resonance lamp to vary. This affects the extent of reversal and therefore the sensitivity for detection. A minimum one standard deviation of (6.5% was found between calculations and the composite data points if the mole fraction of O atoms in the lamp was 5.2 × 10-4 (i.e., 26% O2 dissociation). The resultant COG is plotted as the solid line in Figure 1. The differences between the new and earlier25 (dashed line) COG are insignificant. Hence, because of the above-mentioned consistencies between calibration techniques, we consider the [O] analysis in the present work to be accurate to within about (10%. Kinetics Experiments. Eighteen kinetics experiments were carried out between 1608 and 2091 K under the conditions shown in Table 3. The experiments were similar to the abovementioned calibration determinations, the only difference being that C2H5I was replaced by CH3I. It was of course necessary to increase [O2] substantially, since reaction 3 is known to be much slower than reaction 6. O atom formation rates must exclusively arise from the reactions of CH3 radicals formed from the fast dissociation of CH3I,33 since under similar circumstances we have already shown that I + O2 is negligibly slow over the time period of the experiments.34 Gases. High-purity He (99.995%), used as the driver gas, was from Air Products and Chemicals, Inc. Scientific grade Kr (99.999%), the diluent gas in reactant mixtures, was from Spectra Gases, Inc. The ∼10 ppm impurities (N2, 2 ppm; O2, 0.5 ppm; Ar, 2 ppm; CO2, 0.5 ppm; H2, 0.5 ppm; CH4, 0.5 ppm; H2O, 0.5 ppm; Xe, 5 ppm; CF4, 0.5 ppm) are all either

Figure 2. Measured O atom profiles for two typical experiments. The conditions for the upper trace are P1) 10.99 Torr, Ms ) 2.960, XCH3I ) 1.592 × 10-6, and XO2 ) 5.822 × 10-2 giving T5 ) 2035 K, F5 ) 2.594 × 1018 molecules cm-3, [CH3I] ) 4.129 × 1012 molecules cm-3, and [O2] ) 1.510 × 1017 molecules cm-3. For the lower trace, the conditions are P1) 10.96 Torr, Ms ) 2.597, XCH3I ) 1.592 × 10-6, and XO2 ) 5.822 × 10-2 giving T5 ) 1608 K, F5 ) 2.290 × 1018 molecules cm-3, [CH3I] ) 3.645 × 1012 molecules cm-3, and [O2] ) 1.333 × 1017 molecules cm-3.

inert or in sufficiently low concentration so as to not perturb O atom profiles. Ultrahigh purity grade He (99.999%) for the resonance lamp and high-purity O2 (99.995%) for the atomic filter were from AGA Gases. Scientific grade H2 (99.9999%) and O2 (99.999%), for reaction mixtures, were obtained from MG Industries. Analytical grade CH3I (99%) and C2H5I (99%), both from Aldrich Chemical Co. Inc., were purified by bulbto-bulb distillation, retaining only the middle thirds. Test gas mixtures were accurately prepared from pressure measurements using a Baratron capacitance manometer and were stored in an all-glass vacuum line. Results With the relatively high levels of O2 used in this work, there is some resonance light absorption at 130 nm by O2. Since this absorption is uniform over the O2 bandwidth and [O2] does not change appreciably during an experiment, any observed transmittance decrease in the absence of CH3I reflects both a total density increase due to vibrational relaxation of O2 at low temperature and/or O atom formation from O2 + M f 2O + M at high temperature. Under nearly identical conditions as the kinetics experiments, we corrected for both effects by carrying out several experiments without added CH3I. The transmittance signals from the blanks were then point-by-point subtracted from the corresponding kinetics runs, leaving only the O atom formation signal from CH3 oxidation. Relaxation time constants decreased with increasing temperature and were at most 250 µs near 1600 K. As expected, O atom formation increased with increasing temperature. However, the combination of both effects generally accounted for less than 20%; i.e., greater than 80% of the transmittance signal was due to O atom formation from CH3 oxidation. Using the corrected signals, (ABS)t was determined in each case and [O]t was calculated from the curve-of-growth shown in Figure 1. Figure 2 shows O atom profile results from two experiments, one at high temperature (2035 K) and the other at low temperature (1608 K). For both cases, [CH3I]0 = 4 × 1012 molecules cm-3, and therefore, the available [CH3] is also 4 × 1012 molecules cm-3. To understand these results and the results

CH3 + O2 Reaction

J. Phys. Chem. A, Vol. 103, No. 30, 1999 5945

TABLE 4: Mechanism Used for Fitting [O] Profiles from the CH3 + O2 Reactiona 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. g

CH3I + Kr f CH3 + I + Kr CH3 + O2 f H2CO + OH CH3 + O2 f CH3O + O CH3O + Kr f H2CO + H + Kr H + O2 f OH + O H2CO + Kr f HCO + H + Kr H2CO + Kr f H2 + CO + Kr HCO + Kr f H + CO + Kr HCO + O2 f HO2 + CO HO2 + Kr f H + O2 + Kr OH + OH f O + H2O OH + H2 f H2O + H OH + O f O2 + H OH + H2CO f HCO + H2O I + O2 f IO + O H2CO + O2 f HCO + HO2

k1 ) 4.36 × 10-9 exp(-19858 K/T)b k2 ) 0 k3 ) fitted k4 ) 6.51 × 1013T-6.65 exp(-16740 K/T)c k5 ) 1.62 × 10-10 exp(-7474 K/T)d k6 ) 1.019 × 10-8 exp(-38706 K/T)e k7 ) 4.658 × 10-9 exp(-32110 K/T)e k8 ) 3.1 × 10-7T-1 exp(-8555 K/T)f k9 ) 1.26 × 10-11 exp(-204 K/T)g k10 ) 2.0 × 10-5T-1.18 exp(-24363 K/T)c k11 ) 7.19 × 10-21T2.7 exp(1251 K/T)d k12 ) 3.56 × 10-16T1.52 exp(-1736 K/T)d k13 ) 5.42 × 10-13T0.375 exp(1112 K/T)d k14 ) 5.70 × 10-15T1.18 exp(225 K/T)h k15 ) 7 × 10-11 exp(-30977 K/T)i k16 ) fitted

a All rate constants are in cm3 molecule-1 s-1. b Reference 33. c References 1 and 35. d Reference 30. e Reference 36. f References 1 and 37. References 1 and 38. h References 35 and 39. i Estimated from endothermicity.

from the other 16 experiments, it is necessary to numerically integrate an appropriate chemical mechanism to predict O atom profiles for the experimental conditions. It should be noted that the mechanism need not be all-inclusive, since the secondary chemistry cannot be highly perturbing because [CH3I]0 is so low. Discussion [O]t profiles of the 18 experiments in Table 3 were simulated using the mechanism of Table 4. However, considering only the initial stages of reaction where [O]t e 1 × 1012 molecules cm-3, initial profiles can be analytically determined if reactions 1, 4, and 5 can be considered to be instantaneous. For the conditions of the experiments, these are good assumptions. It is then easy to show that

2k3[CH3I]0 [O]t = {1 - exp(-(k2 + k3)[O2]t)} (k2 + k3)

(7)

An initial estimate for k3 can be made by expanding the exponential term giving [O]t = 2k3[CH3I]0[O2]t. Hence, k3 = RO/(2[CH3I]0[O2]) where RO is the absolute O atom rate of formation (the initial slope) from profiles as illustrated in Figure 2. This approximate expression shows that no information about the branching ratio between reactions 2 and 3 can be obtained from initial O atom profile measurements alone. Initial values of k3 for all experiments were then determined using this approximate procedure. The results can be expressed in Arrhenius form as k3 = 2.4 × 10-11 exp(-16223 K/T) cm3 molecule-1 s-1. Since CH3 is depleting, this estimate will be a lower limit. Starting with the initial values for k3, [O]t was then simulated with reactions 1-5 in Table 4. Some adjustments in k3 were necessary, but under all conditions, initial profiles could be fitted. However, the simulations at longer times showed substantial negative deviations from experiment, particularly at high temperature. Obviously, reactions 1-5 of Table 4 are insufficient to explain the data, and the mechanism must be expanded. We then attempted to use the full GRI mechanism3 to explain the results. The high-temperature experiment shown in Figure 2 was numerically simulated,40 and the predicted behavior is shown as open squares in Figure 3. Clearly, GRI-Mech 2.11 explains neither the initial rate of formation nor the long-time O atom yield. Similar calculations with the Leeds mechanism41 are shown as the dotted line in Figure 3, and this mechanism

Figure 3. Comparison of the results for the 2035 K experiment of Figure 2 to numerical simulations as described in the text, with predictions using (0) the full GRI-Mech 2.11 mechanism (refs 3 and 40), (‚‚‚) the Leeds mechanism (ref 41), (-) an irreversible 48 step reaction mechanism using some of the GRI rate constants, (- - -) the 16 step mechanism of Table 4 using some of the GRI rate constants, and (-‚-‚-‚-) the 16 step mechanism of Table 4 using rate constants for reactions 2 and 3 from Yu et al. (ref 2).

also fails to reproduce the measurements. These mechanisms and the rate constants used in them are optimized against several experimental test cases and therefore cannot be modified in any way. Hence, neither code can be used for explaining the present data. We reduced the number of reactions considered in these mechanisms. By use of GRI-Mech rate constants, the solid line in Figure 3 is an irreversible 48 step reaction mechanism that includes CH3 pyrolysis chemistry. Clearly, the reduction in mechanism size only affects the predictions at times greater than ∼1.5 ms. We reduced the mechanism even more to that shown in Table 4, and again with some of the GRI-Mech rate constants, the predictions (dashed line) are closely similar to the full GRIMech and reduced 48 step mechanisms. Last, we used the revised rate constants of Yu et al.2 (eqs 4 and 5) for reactions 2 and 3, and the prediction is still disappointing, giving the dashed-dotted line in Figure 3. Their downward revision in both rate constants is still insufficient to explain the present data. It is obvious by inspection of Figure 3 that the initial O atom rate will require a rate constant for reaction 3 of about one-half that of eq 5, and this conclusion is reinforced by comparing our approximate k3 to eq 5. The ratio is ∼0.3 over the present T range. It is also obvious from Figure 3 that the overall yield of O atoms at long times approaches 4[CH3I]0,

5946 J. Phys. Chem. A, Vol. 103, No. 30, 1999 suggesting that all three H atoms in CH3 are somehow released to subsequently form O atoms in reaction 6. Reactions 1-5 in Table 4 only account for a maximum of two O atoms per initial CH3, justifying the expanded 16 reaction mechanism of Table 4. At low temperature, the kinetics are almost completely determined by reactions 1-5 of Table 4; however, as temperature increases, other processes can become important. It is important to assess the importance of reaction 2 (yielding H2CO + OH) to the kinetics. As shown by eq 7, this reaction could have a substantial rate constant at low temperature and not affect the initial O atom rate from reaction 3. However, as temperature increases, the maximum O atom yield from reaction 2, followed by reactions 16, 8, 10, and 5 of Table 4, would be 2 O atoms per initial CH3, and we measure yields approaching 4. With the Table 4 mechanism, inclusion of a substantial rate for reaction 2 that is competitive with our initial k3 severely inhibits the total O yield. Any attempt to change the branching ratio to increase O atom production results in worse fits to the data, particularly at short times. The best fits are obtained with k2 ) 0 as shown in Table 4. The present results strongly suggest that reaction 2 is negligible, in essential agreement with the early assertion of Baldwin and Golden.13 The Table 4 mechanism has then been used to fit the 18 experiments of Table 3. To obtain the high measured yields, we have to postulate a fast direct reaction between H2CO and O2 giving HO2 and HCO as products. In actual fact, the adopted 16 step mechanism is still too detailed for the small initial CH3I used in the study. If reactions 9 and 11-15 are deleted, the [O] profile prediction is only decreased by less than 3%. Subsequent deletion of reactions 6 and 7 increases the prediction by about 12%, indicating some importance for H2CO thermal decomposition. However, the most important subsequent reaction is the postulated reaction (i.e., reaction 16 in Table 4). Hence, the minimum mechanism for explaining the present data is reactions 1, 3-5, 8, 10, and 16 in Table 4. These, added together, give for complete conversion CH3 + 4O2 ) 3OH + 4O + CO. By contrast, substituting reaction 2 for reaction 3, with the other processes contributing, gives CH3 + 3O2 ) 3OH + 2O + CO. Hence, any study using OH-radical detection diagnostics will have difficulty in delineating which process, reaction 2 or 3, is important. However, the present O atom diagnostic method can be used to assess the relative importance. Even though half of the reactions are unimportant, the entire 16 step mechanism in Table 4 was used to fit the data. The important rate constants (i.e., eqs 1, 3-5, 8, 10, and 16) are known with sufficient accuracy with the exception of reactions 3 and 16 in Table 4. Hence, the analysis really requires that both rate constants be iteratively changed until a satisfactory fit is obtained. The fits show (a) that [CH3] is depleted at all temperatures and (b) that reaction 16 is fast enough to slightly contribute to [O]t at relatively short times. Even though reaction 3 is almost isolated, reaction 16 has some effect. We find that initial rate estimates for eq 3 (i.e., k3 = 2.4 × 10-11 exp(-16223 K/T) cm3 molecule-1 s-1) have to be only slightly increased (∼12%). The final values for k3 are listed in Table 3. The values for reaction 16 are substantially less accurate at all temperatures because this process accounts for the additional [O] produced at longer times. The iteratively fitted values for the 18 experiments are listed in Table 3. Over the entire time ranges, the mutually derived values reproduce the O atom profiles to within (10% for the experiments below 2000 K and to (10-20% for those above 2000 K. The fits are satisfactory considering the (10% accuracy of the O atom analysis using

Michael et al.

Figure 4. Comparison of a measured O atom profile for a typical hightemperature experiment to a simulation. The conditions are P1) 10.93 Torr, Ms ) 2.999, XCH3I ) 1.592 × 10-6, and XO2 ) 5.822 × 10-2, giving T5 ) 2091 K, F5 ) 2.599 × 1018 molecules cm-3, [CH3I] ) 4.138 × 1012 molecules cm-3, and [O2] ) 1.513 × 1017 molecules cm-3. The line is a fit using the mechanism of Table 4 together with the rate constants for reactions 3 and 16 given in Table 3.

Figure 5. Arrhenius plot of the data for k3 from Table 3. The line is given by eq 8, and the solid circles are the individual data points.

the COG from Figure 1. An example of a fit at T ) 2091 K is shown in Figure 4. Arrhenius plots for reactions 3 and 16 in Table 4 are shown in Figures 5 and 6, respectively. Arrhenius analyses for the rate constants give

k3 ) (3.90 ( 0.40) × 10-11 exp(-16858 ( 1127 K/T) cm3 molecule-1 s-1 (8) and

k16 ) (4.82 ( 0.72) × 10-8 exp(-28476 ( 1628 K/T) cm3 molecule-1 s-1 (9) and these expressions are plotted along with the data in Figures 5 and 6. Reaction 3 is nearly isolated and can be considered to be a direct determination; however, reaction 16 is secondarily derived and should only be considered to be consistent with the measurements. The present determination for k3 can be compared to earlier studies that are mentioned in the Introduction.2,10,11,14-17,20,21 All results are higher by at least a factor of 2-3. The studies that

CH3 + O2 Reaction

J. Phys. Chem. A, Vol. 103, No. 30, 1999 5947

Figure 6. Arrhenius plot of the data from Table 3 for k16 of Table 4. The line is given by eq 9, and the solid circles are the individual data points. The dashed line is the approximate theoretical result given by eq 14.

are most similar to the present are those by Bhaskaran et al.14 and Klatt et al.17 who used O atom ARAS as the diagnostic. They report respective values

k3 ) 1.16 × 10-11 exp(-12900 K/T) cm3 molecule-1 s-1 (10) and

k3 ) 3.80 × 10-11 exp(-15400 K/T) cm3 molecule-1 s-1 (11) where both are valid over the present temperature range. Between 1600 and 2100 K, eq 10 is higher than eq 8 by a factor of 2-3 whereas eq 11 is higher by ∼2. We note that the present O atom ARAS technique is about 5 times more sensitive than those used earlier, and therefore, secondary reactions were more important in both earlier studies in contrast to the present where all radical-radical and many radical-molecule reactions are negligible. The most recent measurement from Braun-Unkhoff et al.19 is

k3 ) 3.42 × 10-12 exp(-12242 K/T) cm3 molecule-1 s-1 (12) and this result is higher than eq 8 by ∼50% at 1600 K but is lower by ∼25% at 2100 K. Hence, eq 12 is in fairly good agreement with the present result. The new result by Hwang, Ryu, De Witt, and Rabinowitz42 is

k3 ) (2.66 ( 0.78) × 10-11 exp(-15813 ( 587 K/T) cm3 molecule-1 s-1 (13) over the temperature range 1575-1822 K. This expression gives values only 31% and 22% higher than eq 8 at 1600 and 1800 K, respectively, and is therefore in excellent agreement with the present results. There have been several theoretical investigations of reaction 3,2,11,13,15,21,43 the most thorough being that of Yu et al.2 On the basis of ab initio calculations44,45 and derived models, they could explain the value for k3, eq 5, determined in their study. We have not carried out theoretical calculations in this paper because fairly slight adjustments in their molecular structures, collisional efficiency factors, Lennard-Jones parameters, energetics, etc. could easily be made to justify the present value (about onethird of theirs). A more important question is whether reaction 2 has any experimental and/or theoretical support. These issues

have also been discussed.2,11,15,21,43,46,47 From a theoretical point of view, reaction 2 should be important if there is labile channel hopping between CH3O2(2A′′) and CH3O2(2A′). The 2A′′ state correlates directly with reactants, CH3(2A′′) + O2(3Σg ), and products, CH3O(2E) + O(3P), whereas the 2A′ state (estimated to be ∼20 kcal mol-1 above 2A′′)11,44,48 correlates with CH3(2A′′) + O2(1∆g) (i.e., the first excited O2 state) and with 1,3 hydrogen migration to form CH2OOH and, ultimately, H2CO + OH. Since the present work indicates little or no importance for reaction 2, we would conclude that either the crossing point is at higher energy than the direct process, reaction 3, or that state crossing is not labile. To explain the profile results, reaction 16 of Table 4 has to have a large rate constant (eq 9). A relatively fast rate has previously been suggested by workers at lower temperatures between 700 and 900 K.49,50 The question arises as to whether this reaction can be justified theoretically. Preliminary electronic structure calculations51 indicate that H2CO(1A1) + O2(3Σg) correlates directly to HCO(2A′) + HO2(2A′′) through a loose 3A′′ transition state. Presuming that the back process, HCO + HO2, has little or no activation barrier, that for the forward process is then the zero-point-corrected endothermicity for the reaction, i.e., 38.94 kcal mol-1. We note in the preliminary work51 that there are two low-valued bending skeletal rocking frequencies that probably correlate with hindered rotations. We have therefore used these calculations to specify a loose transition state with two free internal rotors. Subsequent conventional transition-state theoretical (CTST) calculations with this model give to within (0.8

kCTST ) 7.785 × 10-17T2.049 exp(-19089 K/T) 16 cm3 molecule-1 s-1 (14) for 300 e T e 2200 K. Equation 14 is plotted in Figure 6 where it is seen to be in fair agreement with the data for reaction 16. Equation 14 also predicts values about 3-4 times greater than those reported by Baldwin et al.50 between 713 and 816 K, and it gives values only about 2.5-5 times greater than GRIMech3,52 over the temperature range 1350-2100 K. Since eq 14 is based on a preliminary unoptimized model, it cannot be considered to be accurate. However, the present demonstration suggests that such a process does have a theoretical basis and should be considered as probable. Last, we are suggesting that there is a fast direct reaction between H2CO and O2, and there is a theoretical basis for this claim. Such a process has already been considered in shock tube H2CO oxidation experiments by Eiteneer et al., who measured CO profiles;52 however, a better demonstration would be to make direct O atom measurements on the H2CO/O2 system. Such experiments are planned in our laboratory in the near future. If fast rates for O atom formation are substantiated, these will be used to directly determine k16, and the preliminary electronic structure calculations51 will be refined and completed. This information will then be used in flexible transition-state theoretical calculations of thermal rate constants, thereby allowing for a direct comparison between experiment and theory. Acknowledgment. The authors thank Drs. J. P. Hessler, A. F. Wagner, and Professor P. J. Ogren for carrying out modeling calculations and valuable discussions. We also thank Drs. D.C. Fang and L. B. Harding for preliminary electronic structure calculations and Drs. S. M. Hwang and M. J. Rabinowitz for providing a prepublication version of their paper. This work was supported by the U.S. Department of Energy, Office of

5948 J. Phys. Chem. A, Vol. 103, No. 30, 1999 Basic Energy Sciences, Division of Chemical Sciences, under Contract No. W-31-109-Eng-38. References and Notes (1) Mallard, W. G.; Westley, F.; Herron, J. T.; Hampson, R. F. NIST Chemical Kinetics Database, version 6.0; NIST Standard Reference Data; NIST: Gaithersburg, MD, 1994. (2) Yu, C.-L.; Wang, C.; Frenklach, M. J. Phys. Chem. 1995, 99, 14377. (3) Bowman, C. T.; Hanson, R. K.; Davidson, D. F.; Gardiner, W. C., Jr.; Lissianski, V.; Smith, G. P.; Golden, D. M.; Frenklach, M.; Goldenberg, M. GRI-Mech 2.11, http://www.me.berkeley.edu/gri_mech/. (4) Dean, A. M.; Kistiakowsky, G. B. J. Chem. Phys. 1971, 54, 1718. (5) Izod, T. P.; Kistiakowsky, G. B.; Matsuda, S. J. Chem. Phys. 1971, 55, 4425. (6) Jachimowski, C. J. Combust. Flame 1974, 23, 233. (7) Tsuboi, T. Jpn. J. Appl. Phys. 1976, 15, 159. (8) Olson, D. B.; Gardiner, W. C., Jr. Combust. Flame 1978, 32, 151. (9) Tabayashi, K.; Bauer, S. H. Combust. Flame 1979, 34, 63. (10) Saito, K.; Ito, R.; Kakumoto, T.; Imamura, A. J. Phys. Chem. 1986, 90, 1422. (11) Zellner, R.; Ewig, F. J. Phys. Chem. 1988, 92, 2971. (12) Bowman, C. T. Fifteenth Symposium (International) on Combustion; The Combustion Institute: Pittsburgh, PA, 1975; p 869. (13) Baldwin, A. C.; Golden, D. M. Chem. Phys. Lett. 1978, 55, 350. (14) Bhaskaran, K. A.; Frank, P.; Just, Th. Shock Tubes and Shock WaVes, Proceedings of the Twelfth International Symposium on Shock WaVes; Lifshitz, A., Rom, R., Eds.; Magnes Press: Jerusalem, 1980; p 503. (15) Teitelboim, M. A.; Vedeneev, V. I.; Goldenberg, N. Y.; Karnaukh, A. A. EleVenth International Symposium on Gas Kinetics; Assissi, Italy, 1990. (16) Wu, C. H.; Lin, C.-Y.; Wang, H.-T.; Lin, M. C. In Current Topics in Shock WaVes; Kim, Y. W., Ed.; Seventeenth International Symposium on Shock Waves and Shock Tubes, AIP Conference Proceedings 208; American Institute of Physics: New York, 1990; p 450. (17) Klatt, M.; Ro¨hrig, M.; Wagner, H. Gg. Ber. Bunsen-Ges. Phys. Chem. 1991, 95, 1163. (18) Grela, M. A.; Amorebieta, V. T.; Colussi, A. J. J. Phys. Chem. 1992, 96, 7013. (19) Braun-Unkhoff, M.; Naumann, C.; Frank, P. In Shock WaVes; Brun, Raymond, Cumitrescu, Lucien, Eds.; Nineteenth International Symposium on Shock Waves; Springer: Berlin, Germany, 1995; Vol. 2, p 203. (20) Brabbs, T. A.; Brokaw, R. S. Fifteenth Symposium (International) on Combustion; The Combustion Institute: Pittsburgh, PA, 1975; p 147. (21) Dean, A. M.; Westmoreland, P. R. Int. J. Chem. Kinet. 1987, 19, 207. (22) Tischer, A. L., IV; Hwang, S. M.; De Witt, K. J.; Rabinowitz, M. J. 1996 Fall Meeting of the Western States Section of the Combustion Institute, Stanford University, Stanford, CA; Paper 96F-124. (23) Michael, J. V. J. Chem. Phys. 1989, 90, 189. (24) Michael, J. V.; Sutherland, J. W. Int. J. Chem. Kinet. 1986, 18, 409. (25) Michael, J. V.; Lim, K. P. J. Chem. Phys. 1992, 97, 3228.

Michael et al. (26) Maki, R. G.; Michael, J. V.; Sutherland, J. W. J. Phys. Chem. 1985, 89, 4815. (27) Pamidimukkala, K. M.; Lifshitz, A.; Skinner, G. B.; Wood, D. R. J. Chem. Phys. 1981, 75, 1116. (28) Kumaran, S. S.; Su, M.-C.; Lim, K. P.; Michael, J. V. TwentySixth Symposium (International) on Combustion; The Combustion Institute: Pittsburgh, PA, 1996; p 605. (29) Herzler, J.; Frank, P. Ber. Bunsen-Ges. Phys. Chem. 1992, 96, 1333. (30) Michael, J. V. Prog. Energy Combust. Sci. 1992, 18, 327. (31) Baulch, D. L.; Duxbury, J.; Grant, S. J.; Montague, D. C. J. Phys. Chem. Ref. Data 1981, 10 (Suppl. 1). (32) Lim, K. P.; Michael, J. V. Twenty-Fifth Symposium (International) on Combustion; The Combustion Institute: Pittsburgh, PA, 1994; p 713. (33) Kumaran, S. S.; Su, M.-C.; Michael, J. V. Int. J. Chem. Kinet. 1997, 29, 535. (34) Kumaran, S. S.; Michael, J. V. Proceedings of the 21st International Symposium on Shock WaVes; Houwing, A. F. P., Ed.; Panther Publishing: Fyshwick, Australia, 1997; p 241. (35) Tsang, W.; Hampson, R. F. J. Phys. Chem. Ref. Data 1987, 15, 1087. (36) Kumaran, S. S.; Carroll, J. J.; Michael, J. V. Twenty-SeVenth Symposium (International) on Combustion; The Combustion Institute: Pittsburgh, PA, 1998; p 125. (37) Timonen, R. S.; Ratajczak, E.; Gutman, D.; Wagner, A. F. J. Phys. Chem. 1987, 91, 5324. (38) Timonen, R. S.; Ratajczak, E.; Gutman, D. J. Phys. Chem. 1988, 92, 651. (39) Baulch, D. L.; Cobos, C. J.; Cox, R. A.; Esser, C.; Frank, P.; Just, Th.; Kerr, J. A.; Pilling, M. J.; Troe, J.; Walker, R. W.; Warnatz, J. J. Phys. Chem. Ref. Data 1992, 21, 411. (40) Hessler, J. P. Unpublished results using GRI-Mech 2.11. (41) Pilling, M. J.; Turanyi, T.; Hughes, K. J.; Clague, A. R. The Leeds methane oxidation mechanism, version 1.3, http://www.chem.leeds.ac.uk/ Combustion/Combustion.html. (42) Hwang, S. M.; Ryu, S.-O.; De Witt, K. J.; Rabinowitz, M. J. J. Phys. Chem. 1999, 103, 5949. (43) Benson, S. W. Thermochemical Kinetics; Wiley: New York, 1976. (44) Walch, S. Chem. Phys. Lett. 1993, 215, 81. (45) Sicilia, E.; DiMaio, F. P.; Russo, N. Chem. Phys. Lett. 1994, 225, 208. (46) Ewig, F.; Rhasa, D.; Zellner, R. Ber. Bunsen-Ges. Phys. Chem. 1987, 91, 708. (47) Hessler, J. P.; Du, H.; Ogren, P. J.; Wagner, A. F. 14th International Gas Kinetics Symposium, Leeds, UK, 1996; Paper F3. (48) Hunziker, H. E.; Wendt, H. R. J. Chem. Phys. 1976, 64, 3488. (49) Baldwin, R. R.; Langford, D.; Matchan, M. J.; Walker, R. W.; Yorke, D. A. Thirteenth Symposium (International) on Combustion; The Combustion Institute: Pittsburgh, PA, 1971; p 251. (50) Baldwin, R. R.; Fuller, A. R.; Longthorn, D.; Walker, R. W. J. Chem. Soc., Faraday Trans. 1 1974, 70, 1257. (51) Fang, D.-C.; Harding, L. B. Private communication, October, 1998. (52) Eiteneer, B.; Yu, C.-L.; Goldenberg, M.; Frenklach, M. J. Phys. Chem. 1998, 102, 5196.